BLOG

Guides and notes on AI LiDAR classification.

Working notes on classification codes, formats, terrain, utility corridors, and the review workflow behind production LiDAR outputs.

Top view of a classified airborne LiDAR block with ground, vegetation, buildings, and utilities in distinct classes. — Every tool below produces some version of this. The differences are where it runs, which classes come pretrained, and how it is licensed.

Sep 7, 20268 MIN READWorkflow

LiDAR classification software compared: desktop, cloud, and hybrid tools

Twelve LiDAR classification tools compared on where they run, pretrained classes, custom training, formats, and licence model, from the vendors' own pages.

Read article

Classified airborne LiDAR block with ground, vegetation, and buildings as distinct classes.

Sep 7, 20266 MIN READClassification

Point cloud classification software: what it does and how to evaluate it

What point cloud classification software does to a LAS file, the four jobs it must do well, and a trial you can run on your own tiles before buying.

Classified transmission corridor with conductors and support structures separated from vegetation.

Sep 7, 20266 MIN READUtilities

Powerline LiDAR deliverables: what a utility actually receives

The classified corridor, conductor and structure vectors, the span table, clearance zones, and the formats a utility expects from a powerline LiDAR survey.

Airborne LiDAR point cloud colored by laser return intensity, from blue at low values to red at high values.

Sep 7, 20267 MIN READFormats

Reading a LAS header: what it tells you before you classify

Header block, record format, scale and offset, coordinate system records, LAZ and COPC markers: a field-by-field read of a LAS file before classification.

Colorized airborne LiDAR point cloud of cliffs and canyon vegetation, individual laser returns visible.

Sep 7, 20266 MIN READWorkflow

LiDAR point density and quality levels: what a classifier needs from your flight

Pulse spacing, pulse density, and the USGS quality levels, read from the classifier's side: which classes survive at 2 points per square metre and which need 8.

Classified LiDAR block viewed for review, with classes shown in distinct colours.

Sep 7, 20265 MIN READWorkflow

Reviewing a classified point cloud before delivery: a checklist

A pre-delivery checklist for classified LiDAR: file integrity, ground surface, class leaks, corridor classes, flags, and what to record for the client.

Hillshaded bare-earth digital terrain model derived from airborne LiDAR, showing hillfort earthworks and field boundaries with vegetation and structures filtered out.

Jul 13, 20268 MIN READWorkflow

LAZ to classified point cloud: every processing step explained

The path from a delivered LAZ file to a checked classified point cloud: the eleven steps, and the check each one needs before the next can start.

Perspective view of an airborne LiDAR block classified into ground, vegetation, buildings, and utilities by color.

Updated Sep 7, 20265 MIN READClassification

How AI LiDAR classification works: deep learning for point clouds

How AI LiDAR classification works: what a deep learning model reads in a point cloud, what pretrained means, and where it can run.

Classified LiDAR point cloud with the ground surface separated from vegetation and structures by color.

Jul 6, 20265 MIN READClassification

How accurate is AI LiDAR classification? Measuring and validating results

Measure AI LiDAR classification accuracy honestly: confusion matrix, producer's and user's accuracy, reference tiles, and real point cloud validation.

Classified airborne LiDAR block of an urban utility corridor, with ground, vegetation, buildings, and conductors separated by color.

Updated Sep 7, 20265 MIN READWorkflow

How to choose LiDAR classification software for airborne and UAV data

Choosing LiDAR classification software for airborne and UAV data: a criteria-based buyer's guide to data type, local vs cloud, pretrained AI, and QC.

VGround ground extraction over varied terrain, with the bare-earth surface separated from non-ground returns.

Jun 30, 20266 MIN READTerrain

LiDAR ground classification algorithms: adaptive TIN, cloth simulation, and deep learning

How ground classification algorithms work: progressive morphological filter, adaptive TIN, cloth simulation, and deep learning for bare earth extraction.

Classified residential LiDAR block with building roofs held as a separate class from ground and vegetation.

Jun 27, 20265 MIN READClassification

Building classification from airborne LiDAR: roofs, planes, and footprints

How building classification from airborne LiDAR works: ASPRS class 6, planar roof extraction, footprints, and the signals that separate roofs from canopy.

Classified LiDAR point cloud over terrain with low, medium, and high vegetation shown as separate color-coded classes.

Jun 23, 20265 MIN READClassification

Vegetation classification in LiDAR: low, medium, and high classes

How LiDAR vegetation classification splits low, medium, and high classes by height above ground, and turns them into a canopy height model.

Perspective view of a classified LiDAR point cloud with ground, structures, and vegetation separated by color.

Jun 16, 20264 MIN READFormats

LAS classification flags and noise points: withheld, overlap, synthetic, key-point

How LAS classification flags (withheld, overlap, synthetic, key-point) and noise classes 7 and 18 travel through a LiDAR pipeline without losing data.

Transmission towers and conductors crossing open country at golden hour.

Jun 9, 20265 MIN READUtilities

Classifying utility corridors in LiDAR: conductors, poles, and towers

How LiDAR corridor classification separates conductors, poles, and towers from ground and vegetation, and what reviewers check before a block ships.

Vecten Desktop running LiDAR classification on a workstation.

Updated Sep 7, 20264 MIN READWorkflow

Desktop or cloud: where LiDAR classification should run

Data movement, cost shape, contract rules, and the review loop: how to decide where LiDAR classification runs.

Colorized bare-earth elevation model of mountainous terrain.

May 26, 20265 MIN READTerrain

Ground classification and DTM extraction from airborne LiDAR

Ground (ASPRS class 2) is the foundation of every terrain product. How DTM extraction, ground filtering, and QC turn LiDAR into a dependable surface.

Top-down view of an airborne LiDAR block colored by return intensity, from low to high.

May 19, 20264 MIN READFormats

LAS vs LAZ vs COPC: choosing the right LiDAR format for classification

LAS, LAZ, and COPC hold the same LiDAR points but are built for different jobs. A practical guide to choosing a format for classification work.

Classified urban LiDAR point cloud with ground, buildings, and vegetation separated by color.

May 12, 20265 MIN READClassification

LiDAR classification classes explained: from ASPRS codes to production outputs

A field guide to ASPRS LiDAR classification classes: from LAS codes to the ground, vegetation, building, water, and utility layers reviewers ship.

Evaluate Vecten Desktop on your own LiDAR blocks.

Classify LAS, LAZ, and COPC blocks on your own workstation and publish review-ready outputs.

Request Early Access